OVERVIEW OF TEST REQUIREMENTS ON HV DC APPARATUS AND

RESULTING IMPACTS ON UHV DC TEST SYSTEMS

Carl-Hendrik Stuckenholz1, Pascal Lequitte1, Michael Gamlin1; Richard Connell2
Haefely Test AG, Switzerland; Haefely Test AG, CH-4052 Basel, Birsstrasse 300, Switzerland
2
Hipotronics Inc., 1650 Route 22 North, Brewster, NY 10509, USA;
Email: stuckenholz.carl-hendrik@haefely.com

Abstract: The growing demand for reliable and environmental friendly energy has led to an increased number of High Voltage DC transmission systems. Researchers and manufacturers worldwide are pushing the
limits further to meet tomorrows demand. At the same pace that the HV DC equipment is advancing, the
testing equipment for HV DC equipment have to advance as well so that manufacturers can prove the performance and quality of their components and researchers have the tools to further investigate the insulation
systems at increased voltage levels. This paper provides a brief overview of the different HV DC test requirements and procedures according to IEC standards as well as Cigre recommendations. The peculiarities
for HV DC cables, DC bushings, converters, etc as well as general purpose testing will be introduced and
discussed. An alternative to the common Greinacher cascade UHV DC generators will be presented in detail.
1.

INTRODUCTION

Today, UHV DC installations become more and

more common to provide reliable and environmental friendly energy all over the world. Installations with voltage levels up to 800 kV are being
planned [1-2] or are being built in all parts of the
world. Distances up to 3700 km and power ratings
up to 3150 MW are being realized. Future
UHV DC installations will push these limits even
further with voltage levels of 1000 kV and above
being considered.
For manufacturers and researchers to be able to
exceed todays limits, test equipment has to stay
ahead. For system voltages exceeding 1000 kV,
test voltage levels for the UHV DC apparatus of
approx. 2000 kV are required.
This paper focuses on UHV DC testing requirements according to todays standards and their
impact on test equipment and gives an overview
of other test requirements. Detailed information
regarding Partial Discharges can be found in [3].
2.
2.1.

WITHSTAND TESTING
Converter Transformers,
Smoothing Reactors and Bushings

The requirements of withstand testing are similar

in respect to duration and pass criteria [4-6]. Withstand voltage is applied for a period of 2 hrs with
positive polarity. The acceptance criteria consist

of a no flashover/failure requirement as well as

PD specification.
Bushings for converter transformers have a 15 %
increased test voltage level compared to converter transformers. There are no discrete test
voltage levels. All test voltages have to be defined
under consideration of the component layout of
the UHV DC system like number of valve bridges.
2.2.

Valves for LCC and VSC

Testing of valves is different for thyristor based

Line Coupled Converter (LCC) valves [7], and
IGBT or Diode based Voltage Sourced Converter
(VSC) valves [8]. All test voltage levels have to be
derived from service conditions, etc.
For LCC valves, valve support d.c. voltage test,
multiple valve unit d.c. voltage test to earth and
valve d.c. voltage test, have to be performed.
These tests follow the same procedure. From
50 % of test level U1min, the voltage is raised
within 10 s to 100 %. After 1 min the voltage level
is reduced to U3h and maintained for 3 h, then it is
reduced to 0. The tests differ with respect to connection of the terminals and ground. All tests are
to be done at both polarities, starting with pos.
polarity. Between the tests, the valve has to rest
for at least 2 h. For this reason these tests are not
considered to be Polarity Reversal (PR) tests.
For VSC valves, testing is more extensive. The
maximum continuous operating duty test

( 30 min) and maximum temporary over-load

operating duty test (10 min) require a test voltage
based on the maximum continuous d.c. voltage.
Switching frequency and modulation pattern are to
be based on service conditions.

influences the field distribution inside the cable.

Depending on cable design and temperature of
insulation, the maximum field strength can be
achieved on the cable sheath under load conditions.

The minimum voltage test ( 10 min) is for proving performance of IGBT valves which derive the
necessary operating power from the voltage between the terminals. Valve support d.c. voltage
test and multiple valve unit d.c. voltage test to
earth follow the same scheme as LCC.

The test duration is significantly extended for Development tests with 30 and 40 LC24 cycles compared to 8 and 12 LC24 during Type Test for LCC
and VSC cables respectively.

2.3.

Power Cables

At present, no mandatory IEC standard covers DC

cables. Only recommendations are available from
CIGRE [9-11]. They differentiate between polymer
cable, typically used up to 250 kV and oilpaper/mass-paper/gas pressure cables up to
800 kV. For polymeric cable the recommendations
differ according to the converter configuration.
The test voltage level is slightly increased for
polymeric cables with 1.85 * U0 compared to
1.8 * U0 for routine and type test and 1.45 * U0 to
1.4 * U0 for after installation test (on-site test). Although test duration per CIGRE recommendation
is 15 min for routine and after installation test, a
considerable amount of time is needed for charging/discharging of the cable so that the DUT can
be exposed to voltages above U0 for several
hours [12].
Development tests apply for polymeric cables,
with a test voltage level of 1.45 * U0. The recommendation respects the intended use of the polymeric cable. When used with VSCs, no PR will
occur during standard operation. For this reason
the number of PR test cycles is evenly distributed
over positive and negative withstand test cycles
so that the total amount of withstand cycles is the
same as with polymeric cables used with LCCs.
Type and development testing is divided into load
cycles. While the test voltage is applied permanently during a load cycle (LC), a current source is
used to heat the cable to steady state conditions.
The cycles can be than described as LC24, LC48
and LCZL and LCHL. Heating for LC24 is 8 h and 16
h of cooling, for LC48 24 h heating and 24 h cooling apply. LCZL is zero load, with no heating at all
and LCHL with permanent heating. The heating

Between load cycle blocks with different polarity it

is recommend to rest the cable for a minimum
period of 8 h for non polymeric cables and 24 h for
polymeric cable with the cable conductor connected to the cable sheath.
The long duration of testing is remarkable. For
polymeric cable in development test, continuous
neg. polarity is applied for as long as 160 days. As
these are recommendations by CIGRE, differing
test conditions can be negotiated.
3. POLARITY REVERSAL TESTING
3.1.

Converter Transformers,
Smoothing Reactors and Bushings

Requirements on PR testing are similar in respect

to duration and pass criteria. The PR cycle consists of 90 min neg., 90 min pos. and 45 min neg.
test voltage. Transition time for reversals is limited
to a maximum of 2 min, shorter times preferred.
The acceptance criteria consist of a no flashover/failure requirement as well as PD specification. In the future, the second neg. polarity might
be extended to 90 min as well.
3.2.

Power Cables

PR testing on cable is done for non-polymeric cables and polymeric cables used with LCCs. Polymeric cables used with VSCs are not exposed to
PR and consequently not tested.
During PR testing, the cable is exposed to LC24
heating/cooling cycles. During this cycle, the polarity is reversed every 4 h for non polymeric and
every 8 h for polymeric cables with an allowed
transition time of 2 min. This procedure is repeated for 10 heat cycles at 1.4 * U0. For polymeric cables 8 heat cycles at 1.45 * U0 are applied for type test and 20 cycles at 1.25 * U0 during development tests

4.
4.1.

WET AND POLLUTION TESTING

Smoothing Reactors

The wet test is applied to the insulators of dry-type

smoothing reactors with the insulators arranged in
service condition. Since the wet test is done primarily on the insulators, an equivalent structure
may be used to achieve the same position as in
service.
The test voltage level is the same as the withstand test voltage level with 1.5 * Udmax, with Udmax
being the specified highest continuous dc voltage
between the line terminals and earth. The test
voltage level is applied for 1 h with the wetting of
the insulator beginning 30 min before the test, no
flashovers are allowed.
4.2.

Bushings

Porcelain housed bushings may need to be tested

according to artificial pollution test methods as
described in [6]. Testing is performed at d.c. withstand test voltage level, typically of negative polarity. Testing is to be done in three consecutive
tests. Outdoor bushings may be even wet tested
for 1 h at 1.25 x Udc.
Horizontal wall bushings may be uneven wet
tested in a position resembling service condition.
The voltage level is 1.25 * Udc. The test voltage is
applied before wetting of the outside part of the
bushing. Duration is until a flashover occurs. Otherwise it shall be maintained for at least 15 min
from start of the rain.
Acceptance criteria are the same for all wet and
pollution tests. In case of a puncture the DUT has
failed. In case of flashover, the test is to be repeated once after a condition check. No additional
flashover is allowed.
4.3.

Isolators

For testing of isolators [14] has to be considered.

Testing is requested according to salt fog test
method and/or the solid layer method.
With a current of 100 mA driven through a resistive load, the ripple factor must be 3 %. Voltage
drops 10 % are acceptable. As per the Technical Report, voltage drops between 10 % and
15 % can be tolerated.

Voltage overshoot shall not exceed 10 %. In case

of flashover between 5 % to 10 % overshoot, the
test is considered invalid and has to be repeated.
Under salt fog test, the DUT is sprayed with a solution of defined salinity. Test duration is 1 h for
salt fog test.
The solid layer method defines a pollution layer to
be put on the DUT. When test voltage level is
reached, even fog is generated to produce a defined wetting rate of the pollution layer. Test lasts
until flashover occurs or the risk of flashover is
negligible. The risk can be assumed by evaluation
of the leakage current.
5.

SUPERIMPOSED VOLTAGE TESTING

5.1.

DC and Impulse Testing

Only cables are tested with impulse voltages superimposed on the DC voltage. Table 1 shows
possible configurations of the combined voltages
and the configurations. The indices used are 1
for lightning and 2 for switching impulse, while
S stands for same and O for opposite polarity.
Table 1: Combined voltage tests
Polarity
DC

Lightning
Impulse
UP1

Switching
Impulse
UP2

Pos.

Pos.

Pos.

UP2S

Neg.

Neg.

-UP1

Neg.

-UP2O

Pos.
Neg.

UP1

Pos.
Neg.

UP2O
-UP2S

Depending on the type of cable and the intended

use, only some of the impulses have to be considered for testing. For non polymeric cables and
polymeric cables for LCC converters this is UP2O;
UP2O; UP1; UP1. For VSC cables UP2S and UP2S
apply additionally. Polymeric cables do not have
to be tested with superimposed lightning impulse
if the cable will not be exposed to lightning due to
installation.
The amplitude of the superimposed impulse depends on the overvoltage the cable can experience. For same polarity of DC voltage and impulse, the maximum voltage is often limited by
surge arresters installed in the converter station
while the peak amplitude for opposite polarity may
be limited by the converter itself. Testing is to be
done with at least 1.15 * the highest possible amplitude.

5.2.

DC and AC Testing

Valves for VSC converters have to be tested with

DC and AC test voltage superimposed. Testing is
split into a short term (10 s) and long term
(30 min) test. For short term testing, an energy
storage capacitor might be sufficient. Eventually
this can be the inherent capacitance of the DUT
itself. For long term testing a separate DC power
source is mandatory.
The test levels are defined by the valve and converter station design. The voltage is brought up to
the 10 s test level, reduced to the 30 min test level
and is switched off.
6.

IMPACT ON TEST EQUIPMENT

Testing of DC equipment is very demanding and

very special compared to AC or impulse testing.
The following points have to be considered:
6.1.

Long term testing

Typical test systems are designed for short term

testing, i.e. some hours. For non polymeric DC
cables, test duration is up to 10 days continuous
at one polarity and 30 days of voltage with polarity
changes, rest times are not considered. For development testing of polymeric cable the durations
are as high as 160 days of continuous negative
test voltage, the complete development test procedure lasts for a minimum of 360 days with only
very short interruptions on the test voltage level.
Thermal design: Test equipment is designed for
application testing under consideration of test durations and cycles as required by standard and
customer request. Keeping the on-time to a minimum gives the advantage of higher output ratings
or more compact designs at reduced costs.
Electrical design: For shorter testing durations,
higher electrical stress can be accepted to reduce
size, weight and cost of the system.
Controls and regulators: According to [15], the
test voltage has to be within 1 % for test durations below 60 s and 3 % for test durations exceeding 60 s. As changes of the input voltage are
typically rather slow within some hours, over a
testing period of 160 days it can be assumed that
the input voltage will vary within the 10 % limit
for the power grid. The regulator of the test

equipment has to compensate for this deviation or

additional measures to stabilize the input voltage
have to be taken.
6.2.

High current testing

For DUTs with large capacitances, the charging

current has to be sufficiently high to reach the test
voltage in a reasonable time. Considering 2 min
for PR, charging time has to be less than 1 min.
As the DUTs have capacitive characteristic, the
charging current decreases and should reach
steady state current within this 1 min. While PR is
also required for type and development tests on
dc cables, especially cable parts, routine tests and
after installation tests are done on the complete
cable installation with length well above 100 km.
Even with a high current power supply, charging
can take several hours under these conditions.
The DC equipment must also be compact and
mobile to bring it on-site especially for after installation tests.
For testing insulators, electrodes, etc. pollution
and wet tests maybe required. Because of the
reduced insulation resistance and discharges,
power sources capable of large currents are necessary to reduce feedback effects on the discharges. Currents of up to 2 Amperes are considered to be necessary [13].
6.3.

Stored energy

As mentioned above, time for PR is limited to

2 min for most applications. As DC testing involves charging of large capacitances, they also
have to be discharged in a reasonable time. This
is true not only for standard testing like PR, but
also in case of emergencies like failures of DUT,
testing equipment, etc. and for undefined conditions like power outages.
With impulse generators and AC systems earthing
of the complete system is rather easy. Under DC
conditions, earthing of a fully charged DUT might
cause arcing over a considerable time endangering not only the test equipment and the DUT but
also personnel operating the system. For this reason, additional bleeding resistors might be necessary for testing DUTs with large capacitances.
They help to discharge the DUT to safe levels
within acceptable time. The bleeding resistor will
increase the leakage current making a power

source with higher current rating, for charging and

steady state current, necessary.
6.4.

Long term stability

The considerations on thermal and electrical design for long term testing also apply for the measuring equipment. Stressing the equipment with
heat and electrical field over a long time will affect
the aging of the components and impact the long
term stability of the measuring device.
Long term stability can be estimated from experience and material parameters of the used components, mainly capacitors and resistors for the
high voltage dividers and the low arm.
6.5.

Ripple measurement

Although the ripple factor is as important for describing a DC voltage as the frequency is for AC
voltages, its measurement was often not realized.
This is due to the fact that a separate measurement system had to be used for the ripple measurement. For most DC generators, the ripple frequency is the power frequency or a multiple of it.
In praxis, the limit requires a bandwidth of approx.
7 times of the ripple frequency. For Greinacher
cascades the bandwidth must be 350/420 Hz and
700/840 Hz for symmetrical voltage doubler circuits for power supplies with 50/60 Hz mains frequency.
HVDC dividers realized as resistive dividers have
very limited bandwidth and may produce unacceptable measuring errors for frequencies of a few
Hz already: Capacitive compensated dividers are
mandatory for ripple measurement.
Modern digitizers provide sufficient resolution and
capabilities to measure the arithmetic mean value
and ripple with the required uncertainties within
one measurement device.
7.

DESIGN OF UHV DC TEST EQUIPMENT

Generation of HVDC voltages for test purposes is

well known for several decades. Typical designs
implement a Greinacher cascade. Here, a different approach is presented.
Our modular DC generators were originally designed for powering industrial electron beam systems which are operated 24 hours per day. The
main requirements for these applications include

uninterrupted operation in the presence of continuous flashovers and a mechanically compact

package. Additionally, very high voltages are required with stable voltage regulation and small
voltage drop during transient or steady state
changes in load impedance. These requirements
are similar to those required for HV testing and a
line of modular test systems has been developed
employing the best features of the industrial designs. Systems up to 1800 kV have been delivered and installed to date. A system with 2200 kV
and 30 mA continuous current is projected. For
increased current ratings, the modules can be
configured in parallel. 200 mA short term and
150 mA continuous current have been realized for
a 1000 kV outdoor system.
Up to date, the design has been applied to a
1 stage 3 phase system. The current rating is 2 A
at an output voltage of 300 kV for pollution and
icing testing.
7.1.

Theory of Operation

Fig. 1 shows the electrical diagram for a 1 phase

200 kV single stage DC power supply.
To upper stages
Y1

HV output

Y2
H1
D1

D2

CD1

RB1

CD2

RB2

RM

CM

H2

X1
X2
Mains input

M
DC return to current meter

To kilovoltmeter

Polarity motor

Fig. 1: 200 kV single stage doubler circuit

Each stage of the system has a transformer feeding a symmetrical voltage doubler circuit. The tertiary winding on each transformer supplies voltage
and power to upper stages in the cascade. The
HV transformer core is electrically connected to
the midpoint of the DC voltage doubler capacitors.
The H2 terminal of the step up transformer is also
connected to the mid DC potential point. The result of this connection technique is the establishment of a uniform DC voltage division in the HV
transformer while maintaining a compact design.
The diodes are connected on a rotary switch
which mechanically enables the system to change
from positive to ground to negative polarity within

seconds. The control will reduce the grounding

and reversing speed for normal test conditions to
reduce stress on the switch. For easy handling,
modules rated 400 kV and 600 kV are available.
The height for a 600 kV module is approx. 2.8 m.
The compensated divider is located inside the
module for the measurement of the arithmetic
mean value and the ripple factor of the DC voltage. The divider can be connected/disconnected
on the outside of the module for calibration purposes.
Fig. 2 shows a 1000 kV DC system with a 15 min
current rating of 45 mA and a continuous current
rating of 10 mA for testing of converter transformers. The system consists of the damping resistor
(A), the DC cascade (B), the bleeding resistor (C)
and a coupling capacitor (D) for PD measurement.
The bleeding resistor is not necessary in most
applications. While it supports the damping resistor and makes connections of the DUT a little easier, its main purpose is to fast discharge the capacitance of the system of converter transformer
and coupling capacitor for PR times in less than
2 min. The damping resistor protects the DC cascade in case of failure or flashover by limiting the
current through the cascade and the diodes. It
acts also as a filter for blocking the switching
noise of the diodes for PD measurements.

design. Fig. 3 visualizes the space requirements

for a classic Greinacher cascade and the modular
design with comparable technical specification.

Height [m]
top electrode [m]
Voltage
[kV]
@ ripple < 3 %
Current [mA]

Greinacher
Cascade
12
9.7
1500
20 cont.

Modular
Cascade
11.2
3.1
1600
1800 (@ 5 mA)
10 cont. / 20 (4 hr ON)

Fig. 3: Greinacher cascade 1500 kV 20 mA left

Modular cascade 1600 kV 20 mA right
Thanks to this design, it is easy to move the test
system in the test lab with help of air cushions.
The compact modular design allows also easy onsite testing under fair weather, e.g. of cables, with
very short set-up time. Generators with outdoor
design are also available.
Expansion to higher voltages can be done by
simply adding modules to an existing system,
without reducing the current rating of this system.

A
B
C
D

With an additional front end, it is even possible to

split a system into two independent test systems,
for example for bipolar operations.
Dust is a major problem for DC test equipments.
As the active parts of our generators are inside a
limited number of FRP cylinders, the test system
is easy to clean and requires less maintenance
than a Greinacher cascade.

Fig. 2: UHV DC system during commissioning

High outputs currents can be achieved by parallel

connection of modules.

7.2.

7.3.

Advantages

The generator, as well as the coupling capacitor

and the bleeding resistor, is made of stacked FRP
cylinders. Together with the compactness of the
generator modules, it leads to a light and modular

Influences on Voltage Stability

The changes in mains voltage, usually less than

10% and rather slow, cause the DC output voltage
to change proportionally. The control unit keeps
the test voltage within 1 % of the target value. In

case of an instantaneous change in the mains

voltage of 10 %, the automatic voltage control will
readjust the output voltage in approximately 1 s.
When a test object is subjected to DC voltage, the
equivalent load resistance can change due to
heating effects, dielectric absorption, glow corona
discharges or other phenomena. When the DC
output current changes instantaneously from noload to full resistive value, the output voltage will
drop linearly as long as the mains voltage is fixed.
The typical value of the voltage drop from no load
to full load current is < 15 %. During testing, the
transition between no load and full load appears
rather slow and will be compensated by the static
control to 1 % of target value.

9.

[4]

IEC 61378-2 ed. 1.0 (2001-02), Part 2

Corona, surface discharges, intense streamers,

etc, cause transient changes in the load resistance and capacitance. Due to system limitations
in power supply and regulation, these transients
cannot be compensated by the control and voltage regulator. Their influence can be reduced with
systems with higher current rating or by increasing
the HV capacitance.

[5]

IEC 60076-6 ed 1.0 (2007-12), Part 6

[6]

IEC 62199 ed. 1.0 (2004-05)

[7]

IEC 60700-1 ed 1.2 (2008-11)

[8]

IEC 62501 ed. 1.0 (2009-06)

The voltage drop can be approximated by:

peak DC current time duration
UDC
" Capacitanc e" of DC generator
For example, with a 1200 kV generator for an output voltage drop of < 10 %, a transient streamer
drawing a current of 100 mA can burn up to 2 ms
before the voltage drop exceeds 10 %. With an
increased capacitance the arc could last longer.

8.

CONCLUSION

The current standards for HV DC testing have

been analysed and requirements for testing
equipment have been derived. An additional solution for UHV DC generators to the common Greinacher cascade has been presented.
Although generation of DC voltage is well known
for decades, development is still ongoing. Recently, the measurement of the ripple factor has
been implemented into the system without the
need for additional external measuring equipment.
Further improvements will be done to reduce the
PD level of the DC cascades. Another area of research is to increase the current capabilities of the
generator while keeping the compact, modular
design with all its advantages.

REFERENCES

[1]

IEC SMB/4272/R, TC 115: Strategic business plan, High Voltage direct Current Transmission for DC voltages above 100 kV

[2]

Husler, M., et al: From Smart Grid to Super

Grid - Efficient Solutions with UHV DC Bulk
Power Transmission of Green Energy, Isoliersysteme bei Gleich- und Mischfeldbeanspruchung, ETG Fachbericht 125, VDE Verlag

[3]

Stuckenholz, C.-H. et. al.: PD requirements

for testing HVDC equipment and their impact
th
on HVDC test equipment; 17 ISH 2011

[9]

Cigre WG 21.01: Recommendations for tests

of power transmission DC cables for a rated
voltage up to 800 kV, Electra 189, 2000-04

[10] Cigre TF B1.16: Add. to Recommendations

for tests of power transmission DC cables for
a rated voltage up to 800 kV, Electra 218
[11] Cigre WG 21.01: Recommendations for testing DC extruded cable systems for power
transmission at a rated voltage up to 250 kV,
Technical Brochure 219, 2003-02
[12] Worzyk, T.: Submarine
Springer Verlag Berlin 2009

Power

[13] IEC 1245 ed. 1.0 (1993-10)

[14] IEC 60060-1 ed. 3 (2010-09), Part 1
[15] IEC 60060-2 ed. 3 (2010-11), Part 2

Cables;